Genetic susceptibility to Tardive Dyskinesia in chronic schizophrenia subjects: V. Association of CYP1A2 1545 C>T polymorphism

ORIGINAL ARTICLE

Genetic susceptibility to Tardive Dyskinesia in

chronic schizophrenia subjects: I. Association

of CYP1A2 gene polymorphism

AK Tiwari1

SN Deshpande2,3

AR Rao4

T Bhatia3

SR Mukit3

V Shriharsh3

B Lerer5

VL Nimagaonkar3,6

BK Thelma1,3

1Department of Genetics, University of DelhiSouth Campus, New Delhi, India; 2Department ofPsychiatry, Dr RML Hospital, New Delhi, India;3Indo-US Project on Schizophrenia Genetics,New Delhi, India; 4Biometrics Division, IndianAgricultural Statistics Research Institute,New Delhi, India; 5Department of Psychiatry,Hadassah-Hebrew University Medical Center,Ein Karem, Jerusalem, Israel; 6Department ofPsychiatry, University of Pittsburgh, PA, USA

Correspondence:Professor BK Thelma, Department ofGenetics, University of Delhi South Campus,Benito Juarez Road, New Delhi 110021,India.Tel: þ91 11 24678201Fax: þ91 11 26885270E-mail: [email protected]

Received: 06 January 2004Revised: 15 August 2004Accepted: 06 September 2004Published online 26 October 2004

ABSTRACTUnderstanding the pharmacogenetic basis of developing iatrogenic disorderssuch as Tardive Dyskinesia (TD) has significant clinical implications. CYP1A2,an inducible gene of the cytochrome P450 family of genes, has beensuggested to contribute to the metabolism of typical antipsychotics insubjects with schizophrenia on long-term treatment, and has beenconsidered as a potential candidate gene for development of TD. In thisstudy, we have investigated the significance of CYP1A2 gene polymorphismsin TD susceptibility among chronic schizophrenia sufferers (n¼335) fromnorth India. TD was diagnosed in B29% (96/335) of these subjects. Of the96 TD positives, 28 had been treated with typical antipsychotics alone, 23with atypical antipsychotics alone and 45 patients had received both classesof drugs during the course of their illness. Out of the six SNPs tested,CYP1A2*2, *4, *5, *6 were found to be monomorphic in our population.CYP1A2*1C and CYP1A2*1F were polymorphic and were analyzed in thestudy sample. Since these two allelic variants lead to lesser inducibility amongsmokers, the smoking status of TD patients was also considered for allsubsequent analysis. We observed increased severity of TD among TD-Ysmokers, who were carriers of CYP1A2*1C (G4A) variant allele and hadreceived only typical antipsychotic drugs (F(1,8)¼9.203, P¼0.016). Nosignificant association of CYP1A2*1F with TD was observed irrespective of theclass of drug they received or their smoking status. However, we found asignificant association of CYP1A2*1F with schizophrenia (w2¼6.572, df¼2,P¼0.037).The Pharmacogenomics Journal (2005) 5, 60–69. doi:10.1038/sj.tpj.6500282Published online 26 October 2004

Keywords: north India; Tardive Dyskinesia; CYP1A2; single-nucleotide polymorph-isms; association; pharmacogenetics

INTRODUCTIONTardive Dyskinesia (TD) is a movement disorder characterized by choreoathetoticmovements, and affects B20% of schizophrenia sufferers on long-term treatmentwith typical antipsychotics. A genetic predisposition for developing TD amongpersons suffering from schizophrenia has been observed based on family andtwin studies.1,2 Advancing age, duration of treatment, female gender and organicbrain abnormalities are other well-known risk factors.3,4 Several theorieshave been postulated to explain the pathophysiology of TD, the dopamine

The Pharmacogenomics Journal (2005) 5, 60–69& 2005 Nature Publishing Group All rights reserved 1470-269X/05 $30.00

www.nature.com/tpj

overactivity hypothesis being one of them.5 This hypothesishas received considerable support from pharmacological aswell as physiological evidences.6–9 Efforts to elucidate thegenetic etiology of TD have been made. For example, a largenumber of association studies to evaluate polymorphisms ingenes from the neurotransmitter pathway, including dopa-mine receptors (D2 and D3), serotonin receptors (HTR2A,HTR2C) and the serotonin transporter (5-HTT) have beencarried out. Results from these studies have been incon-sistent or untested in additional samples (for review seeBasile et al10) with the possible exception of the DRD3ser9gly polymorphism, which has been supported by apooled and meta-analysis.11 Thus, pathophysiological me-chanisms underlying TD remain poorly understood. Asecond large group of genes extensively investigated forassociation with TD are the cytochrome P450 (CYP450)family of genes.

The CYP450 enzymes are a highly diverse group of hemecontaining monooxygenases catalyzing a variety of reac-tions, the most common being hydroxylation. The CYPscatalyze phase I metabolism of drugs, that is elimination ofthe drug from the circulation, thus influencing the plasmalevel of the drug. The other factors that affect the plasmalevel of the drug include absorption from the site ofadministration and distribution in the body. The CYP1,2 and 3 family of genes is responsible for metabolism of amajority of xenobiotics. CYP3A4 is responsible for theelimination of B50% of commonly prescribed antipsycho-tics and antidepressants; CYP2D6 metabolizes approximately30% and the rest is accounted for by CYP1A2, CYP2C9,CYP2C19 and CYP2E1.12 Considerable interindividual andinterethnic variation in the extent of drug metabolism hasbeen observed. This variation in drug clearance maycontribute to the pharmacogenetic differences observedamong different patients/populations. Such variation mayalso explain differential predisposition to iatrogenic disor-ders such as TD. Associations of the allelic variants causingthe poor metabolizer phenotype of CYP2D6 with TD havebeen reported in several populations, suggesting a possiblerole of drug disposition in TD susceptibility.10,13,14

CYP1A2 is involved in the metabolism of atypicalantipsychotics such as clozapine and olanzapine. It has alsobeen suggested to contribute to the metabolism of typicalantipsychotics in subjects with schizophrenia on long-term treatment, and has therefore been considered as apotential candidate gene for susceptibility of TD. ThoughCYP1A2 is considered to be a low-affinity high capacitymetabolic clearance pathway for typical antipsychoticscompared to CYP2D6, saturation of the latter duringchronic long-term treatment makes CYP1A2 an importantenzyme for antipsychotic metabolism.15 Smoking is a potentinducer of CYP1A2 gene as is evident from the increased rateof caffeine biotransformation. A higher prevalence ofsmoking B80%16 has been reported in schizophreniasubjects compared to B25% in the general population.Thus, subjects who are smokers would be expected to have ahigher rate of CYP1A2-mediated drug disposition as com-pared to nonsmokers.

In total, 12 different single-nucleotide polymorphisms (SNPs)in CYP1A2 have been reported (www.imm.ki.se/CYPalleles).However, only two of these SNPs namely CYP1A2*1C (G4A)and CYP1A2*1F (C4A) have been shown to be of functionalsignificance.17,18 The CYP1A2*1C (G4A) polymorphism inthe upstream region of the human CYP1A2 gene (�3858)results in a significantly reduced induction of this geneamong smokers (measured by the rate of caffeine 3-demethylation17). CYP1A2*1F (C4A), an intronic poly-morphism was observed in one out of eight healthyvolunteers.18 Thereafter, PCR-RFLP analysis of this poly-morphism among nonsmokers and smokers revealed 24subjects with the CC genotype (10%), 104 with AC genotype(44%) and 108 subjects homozygous for the AA genotype(46%) were identified. A 1.6-fold higher metabolic activitywas observed among smokers homozygous for A allelecompared with the other genotypes in this category.Nonsmoking subjects did not show any significant differ-ences in the metabolic activity. In view of the above results,Basile et al (2000) investigated whether this mutationcontributes to the risk of developing TD in chronicschizophrenia subjects. Abnormal Involuntary MovementScale (AIMS19) scores were used to assess the severity of TDamong 85 schizophrenia subjects. A significantly high meanAIMS score was found among the patients with the CCgenotype, 2.7-fold greater than individuals with the ACgenotype and 3.40-fold greater than AA genotype. Amongsmokers homozygous for the CC genotype, a higher meanAIMS score was observed compared to smokers with AC andAA genotypes (4.7- and 5.4-fold greater, respectively). Theseresults lend considerable support to the existing hypothesisthat the rate of clearance of the drug from circulation iscorrelated with the risk for developing TD. However, a studyon a group of German subjects with schizophrenia did notobserve a significant difference in mean AIMS score betweenthe three genotypic classes.20 In a second study in subjects ofChinese descent, no association between the C4A poly-morphism of CYP1A2 and TD was observed.21 Though aconsiderable number of association studies on other candi-date genes have been reported in different populations,CYP1A2 polymorphisms have not been investigated exten-sively in other populations. Considering the high pharma-cokinetic significance of this gene in the development of adebilitating disorder such as TD we have analyzed thepotential contribution of SNPs (individually or in conjunc-tion) in CYP1A2 gene, to the risk of developing neurolepticinduced TD among schizophrenia subjects from north Indiausing a case–control approach. As a prerequisite, SNPs werevalidated in our population in 250 healthy individuals. Thisis the first report of association study of CYP450 poly-morphisms with TD among Indians.

RESULTSOf the 335 subjects with schizophrenia, 96 (28.66%) werediagnosed to have TD based on the AIMS scale. Thedemographic characteristics of this population are presentedin Table 1. Mean AIMS score of the subjects with TD was

Association of CYP1A2 polymorphisms with Tardive DyskinesiaAK Tiwari et al

61

www.nature.com/tpj

6.5273.482. No statistically significant difference was foundin the distribution of males and females between cases andcontrols or between subjects with TD (TD-Y) and without TD(TD-N). In the TD-Y group, males had higher mean AIMSscore as compared to the females (F(1,93)¼4.305, P¼0.041,r2¼0.222, power¼ 0.537). The severity of TD increased withage (Pearson r¼0.365, P¼0.001) as has been reported inother populations.

Of 96 TD-Y cases, 28 individuals were on typicalantipsychotic medication (mean AIMS score 6.7173.68;duration of schizophrenia 12.78710.60 years), 23 hadreceived only atypical antipsychotics (mean AIMS score5.6973.24; duration of schizophrenia 6.8775.20 years) and45 had received both typical and atypical drugs at some timeduring the course of their illness (mean AIMS score6.4073.44; duration of schizophrenia 10.6978.76 years).There was no significant difference in mean AIMS scoresbetween the three categories (F(2,92)¼0.88, P¼0.916).Although these data on the class of drugs that the subjectsreceived was available, no reliable information on the drugdosage could be documented.

The prevalence of smoking in subjects with schizophreniawas 23.51%. Since the allelic variants (CYP1A2*1C andCYP1A2*1F) lead to lesser inducibility among smokers, theTD-Y group n¼96 was further subdivided into smokers(n¼31) and nonsmokers (n¼ 65) for all further analysis.Since the subjects in this study had taken either typical oratypical or both classes of drugs, we have analyzed subjectsin each of these groups separately. Thus, we have threedistinct patient groups (TD-Y vs TD-N; TD-Y subjectssegregated according to the three classes of drugs; and TD-Ysmokers only with three classes of drugs, for all statisticalanalysis).

The baseline allele frequency for each of the six poly-morphisms was established by genotyping 250 controls. Ofthe six tested SNPs, CYP1A2*2, CYP1A2*4, CYP1A2*5,CYP1A2*6 did not show any polymorphism in our studysample. The absence of polymorphism at these four loci inour samples was confirmed by randomly sequencing five

samples for each amplicon (data not shown). CYP1A2*1Cand CYP1A2*1F were polymorphic and were analyzedfurther. Their genotypic distributions among the studysamples are given in Tables 3 and 4.

CYP1A2*1CThe allele frequencies for CYP1A2*1C in the controlpopulation is G¼ 0.937, A¼0.0612 (PIC¼0.1763). Thedistribution of the genotypes in the control populationwas not in Hardy–Weinberg equilibrium (HWE) (w2¼5.33,df¼1, P¼0.0250). However, the observed genotypic fre-quencies did not deviate from those expected under HWEfor subjects with and without TD and on different drugregimes (P40.05). There was no statistically significantdifference between the distribution of genotypic frequenciesbetween the TD-Y vs TD-N group and TD-Y vs TD-Ncategorized based on the different class of drugs (P40.05;Tables 2 and 3).

The mean AIMS scores among TD-Y subjects who weresmokers with different genotypes were compared. No 2-2homozygotes were observed among smokers. Among theTD-Y subjects who had been on typical antipsychotics andwere smokers, the mean AIMS score was significantly higherfor individuals heterozygous for the variant allele(13.0071.414 vs 7.5674.22; P¼0.016; Table 4). Since thenumber of samples under each genotype category wasnegligible for TD-Y smokers under the atypical drugtreatment ANCOVA was not performed. When we consid-ered the subjects who received typical drugs along withthose who received both classes of drugs, no significantdifference was observed (P¼0.570; Table 4).

CYP1A2 is also inducible by a variety of other environ-mental agents apart from smoking. We analyzed whetherthis polymorphism has an overall significance in TD per se.Among subjects who received typical antipsychotic drugs,heterozygous carriers of the variant allele (1–2) had a slightlyhigher mean AIMS score (9.7573.86 vs 6.2173.48) and wassignificantly different from the homozygous wild type (1–1;

Table 1 Demographic characteristics of the population under study

Schizophrenia subjects Controls P-value

TD-Y TD-N P-value Total

Age (mean7SD) 34.53712.6 31.42710.2 0.014S 32.42711.10 34.05712.78 0.110NS

Sex Male 59 123 0.097NS 182 140 0.167NS

Female 37 116 153 103Smoking status Smokers 31 48 79 NA

Nonsmokers 65 191 256 NADrug typea Typical 28 82

Atypical 23 57Both 45 106

TD-Y¼patients with TD; TD-N¼patients without TD.

The distribution of age among Schizophrenia subjects vs controls and TD-Y vs TD-N was tested using student’s t-test. w2 test was used to compare the sex ratio.

S¼ significant; NS¼not significant; NA¼not available.aSubjects classified according to the class of drugs they had taken either atypical or atypical anitpsychotics or both the drug types during the course of illness.


62

The Pharmacogenomics Journal

P¼0.030; Table 5). However, no significant difference in themean AIMS score was observed when subjects in the typicalantipsychotic group and those who received both classes ofdrugs were combined (P¼0.166). Similarly, the mean AIMSscore was not significant for the subjects who had receivedonly atypical drugs across the two genotypic categories(P¼0.318; Table 5).

CYP1A2*1FThe baseline allele frequency of CYP1A2*1F (PIC¼0.4998)in the control population studied was C¼0.488 andA¼0.512. The genotypes were in HWE in the controlpopulation (w2¼ 3.4885, df¼1, P¼ 0.10). The observedgenotypic frequencies did not deviate from those expectedunder HWE for subjects with and without TD and those ondifferent drug regimes (P40.05). The distribution of geno-typic frequencies between the TD-Y vs TD-N group and TD-Y

vs TD-N categorized based on different classes of drugs wasnot statistically significant (P40.05; Tables 2 and 3).

A comparative analysis of severity of TD among smokerswith different genotypes was also carried out. There was nosignificant association between the allelic state of theindividual and the AIMS score for the subjects who weresmokers and were under typical antipsychotic medication(P¼0.101; Table 4). The SNP was not significantly associatedwith TD even when the entire group (all subjects receivingtypical antipsychotics) was considered (P¼0.425; Table 5).Similar nonsignificant results were obtained when thesubjects who received typical antipsychotics were consid-ered along with those who received both classes of drugs(P¼0.737; Table 5) as well as for subjects who had receivedonly atypical drugs (P¼ 0.929; Table 5). Haplotype frequen-cies of CYP1A2*1C and CYP1A2*1F (for TD-Y and TD-Nsubjects who received typical or atypical drugs) were

Table 3 Genotype distribution among patients with TD (TD-Y) and without TD (TD-N) under different drug usage

Class of drug Status Genotype distribution P-value Power

1\1 1\2 2\2

CYP1A2*1Ca Typical TD-Y 24 4 0 w2¼0.00, df¼1, 0.0722TD-N 60 12 1 P¼1.00b

Atypical TD-Y 21 2 0 w2¼0.041, df¼1, 0.0196TD-N 45 2 0 P¼0.839

Typical+both TD-Y 64 7 0 w2¼1.78, df¼1, 0.2135TD-N 147 28 1 P¼0.182b

CYP1A2*1Fc Typical TD-Y 12 9 3 w2¼2.27, df¼1, 0.1953TD-N 24 37 12 P¼0.132b

Atypical TD-Y 6 12 4 w2¼0.07, df¼1, —d

TD-N 13 24 9 P¼0.932b

Typical+both TD-Y 24 27 12 w2¼0.449, df¼2, —d

TD-N 62 84 30 P¼0.799

aAllele1 is G, allele 2 is A.

bCells with small values (2/2) were merged with the heterozygous group(1/2).cAllele1 is A, allele 2 is C.dThe difference in allele frequencies is negligible and therefore no power.

Yates correction has been applied if the cell values were less than five.

Table 2 Genotype distribution among patients with TD (TD-Y) and without TD (TD-N)

TD-Y TD-N

Allele frequency Genotype distribution (%) Allele frequency Genotype distribution (%)

1 2 1\1 1\2 2\2 1 2 1\1 1\2 2\2

CYP1A2*1Ca 181(0.953)

9(0.047)

86(90.52)

9(9.47)

0(0)

414(0.928)

32(0.072)

192(86.10)

30(13.45)

1(0.45)

CYP1A2*1Fb 100(0.5814)

72(0.4186)

30(34.88)

40(46.51)

16(18.6)

258(0.581)

186(0.4189)

75(33.78)

108(48.65)

39(17.57)

aAllele 1 is G, allele 2 is A.

bAllele 1 is A, allele 2 is C.

Power for genotypic test for CYP1A2*1C is 0.1670; for CYP1A2*1F is 0.05.


63

www.nature.com/tpj

calculated and compared. There was no significant associa-tion of any specific haplotype with TD (P40.05).

We also investigated whether the above two polymorph-isms are a risk factor for the development of schizophrenia.It was found that the genotypic distribution of CYP1A2*1Cis not statistically significant between the controls (age andsex matched) and subjects with schizophrenia (w2¼2.405,df¼2, P¼0.301). Genotypic distribution of CYP1A2*1F washowever statistically significant at 5% level (w2¼6.572,df¼2, P¼0.037; Table 6). Haplotypic frequencies calculatedand compared between cases and controls did not showassociation (P¼0.147). A strong linkage disequilibrium wasobserved between CYP1A2*1C and CYP1A2*1F (D0 ¼0.89;r2¼0.05).

DISCUSSIONThe present study was undertaken to establish association, ifany, of six functional SNPs in CYP1A2 gene with TD, among

schizophrenia subjects from north India. Since CYP1A2 isone of the major genes involved in metabolism ofxenobiotics and TD is an iatrogenic disorder developing inabout 20–30% of all schizophrenia subjects on long-termtreatment with typical antipsychotics, study of SNPs in thisgene has major pharmacogenetic implications. This is thefirst report on both the genetics of TD as well as SNPs inCYP1A2 gene from this ethnically distinct population. Theseverity of TD has also been observed to increase with age inconformity with other reports.4,11,22 However, severity of TDwas higher among males compared to females in our study,which is in contrast to other reports.23,24 This may be due tosampling variation or may reflect a variation unique toIndian population.

Of the 96 TD positive subjects, while 28 had received onlytypical antipsychotics, 23 had received only atypical anti-psychotic drugs. Although the mean AIMS scores betweenthese two groups of TD patients were not significantlydifferent, the duration of schizophrenia in the former group

Table 5 Distribution of mean AIMS scores by genotype in TD positives categorized according to the class of drug(s) taken

Drug type AIMS score (mean7SD) (n) P-value

1\1 1\2 2\2

CYP1A2*1C Typical 6.2173.48 (24) 9.7573.86 (4) — F(2,25)¼5.290, P¼0.030,r2¼0.291, power¼0.599a

Atypical 5.6773.39 (21) 6.0070.00 (2) — w2¼0.997, df¼1, P¼0.318b

Typical+both 6.2973.44 (64) 8.1473.98 (7) — w2¼1.918, df¼1, P¼0.166b

CYP1A2*1F Typical 7.0874.19 (12) 5.4472.60 (9) 6.3373.1 (3) F(2,20)¼0.892, P¼0.425,r2¼0.130, power¼0.182a

Atypical 5.6773.08 (6) 5.5073.61 12) 6.5073.51 (4) F(2,18)¼0.074, P¼0.929,r2¼0.151, power¼0.060a

Typical+both 6.6374.31 (24) 6.3672.80 (27) 5.7573.14 (12) w2¼0.611, df¼2, P¼0.737b

aANCOVA.

bKruksal–Wallis H test.

n¼number of individuals in each category.

Bold values are statistically significant.

Table 4 Distribution of mean AIMS scores by genotypes in TD positive smokers

Drug type AIMS score (mean7SD) (n) ANCOVA

1\1 1\2 2\2

CYP1A2*1C Typical 6.6773.67 (9) 13.071.41 (2) — F(1,8)¼9.203, P¼0.016,r2¼0.666, power¼0.757

Typical+both 8.1773.53 (18) 8.8074.66 (5) — F(1,20)¼0.333, P¼0.570,r2¼0.188, power¼0.085

CYP1A2*1F Typical 11.2572.99 (4) 3.5070.70 (2) 7.0074.24 (2) F(2,4)¼4.285, P¼0.101,r2¼0.712, power¼0.414

Typical+both 9.4374.35 (7) 7.3872.97 (8) 8.0073.464 (3) F(2,14)¼0.685, P¼0.520,r2¼0.155, power¼0.143

n¼number of individuals in each category.

Bold values are statistically significant.


64


was higher. However, we would be cautious in interpretingthese results due to the nonavailability of the drug dosage inour study sample. Our observation of a comparable numberof subjects developing TD while on only atypical antipsy-chotic medication is notably different from most otheravailable reports on drug induced TD, where TD has beenshown to be largely induced by typical antipsychoticdrugs.10 As far as atypical antipsychotic drugs are concerned,weight gain and not TD has been one of the major sideeffects documented.10 Such a difference between most of theother reports and the present study may reflect basic geneticdifference(s) underlying development of TD among differ-ent ethnic groups. Alternately, use of atypical antipsychoticdrugs in most countries may be rather limited due to costand therefore there may have been limited scope to evaluatethis effect.

The absence of polymorphism at four SNPs (*2, *4, *5, *6)in the CYP1A2 gene in our population is different from theavailable report in French population.25 However, it may benoted that the allele frequency of the SNPs were consider-ably low (0.5–1%) even in that population. This mayindicate either a recent evolution of these SNPs in thehuman genome (considering the highly adaptive nature ofdrug metabolism26) or may reflect the highly conservednature of this gene. However, the reported highly variableactivity of this gene (10- to 30-fold27) may be due topolymorphism(s) in the conserved domains (3-Methylcholanthrene elements) in the upstream region of the gene assuggested recently.28 A very low frequency of polymorphismhas been observed in CYP1A2*1C (PIC¼0.1763) in this3-MC element in our study sample. This is similar to theEgyptian population29 but significantly different from thatreported in Japanese population.17 Consequently, absence ofassociation of this SNP with TD in our study seems in order.Since no other reports on association of this SNP with TD areavailable, comparisons are not possible. The severity of TDamong subjects who were smokers as well as treated onlywith typical antipsychotic drugs (n¼ 11) was significantlyhigher in subjects carrying the single copy of the variant(13.0071.414 vs 7.5674.22; P¼ 0.016). This could be

explained on the basis of differential inducibility in thepresence of the variant allele. However, since only two suchheterozygous individuals were observed (Subject-1 AIMSscore¼12; Subject-2 AIMS score¼ 14) we would be cautiousin giving any weightage to this observation. A study havinga larger sample size is warranted for conclusive inference.Increased severity was also observed when the complete TD-Ygroup (both smokers and nonsmokers) treated with onlytypical antipsychotics (n¼28) was considered (9.7573.86 vs6.2173.48; P¼0.030). However, this was contributed by thesame two heterozygous (smokers) individuals mentionedabove.

The allele frequency of CYP1A2*1F in our study sample isslightly different from those reported for Caucasian popula-tions.15,18,29 However, it is similar to the frequency reportedfor African Americans.15 Association of TD with CYP1A2*1Fhad earlier been reported by Basile et al.15 However, Schultzeet al20 failed to replicate this finding. In our population alsowe did not observe any significant association between TDand *1F. The lack of association could be attributed toseveral factors: (a) Differing genetic backgrounds as partlyevident from differing allele frequencies in this gene; (b) TheCYP1A2*1F (and also*1C) gene is inducible by smoking butsmoking was not prevalent among schizophrenia subjects inour samples (however, CYP1A2 is induced by a variety ofother substances such as glucosynolates present in crucifer-ous vegetable and charboiled meat); and (c) The sample sizeis small and thus reduced statistical power to detect anyassociation if the gene is of minor effect (Table 2). However,since the sample characteristics with specific reference toprevalence of smoking were notably different from theprevious reports,15,20 any direct comparison between theirobservation and this study on ethnically distinct Indianpopulation is inappropriate.

The mean AIMS score in TD-Y individuals on medicationonly with typical drugs and who were homozygous for thevariant allele (AA) of *1F was higher in comparison to othergenotypic categories but was not statistically significant(Table 5). Similar trend was observed among smokers ontypical drug treatment only (Table 4). However, as in the

Table 6 Genotype distribution among subjects with schizophrenia and controls

Schizophrenia cases Controls

Allele frequency Genotype distribution (%) Allele frequency Genotype distribution (%)

1 2 1\1 1\2 2\2 1 2 1\1 1\2 2\2

CYP1A2*1Ca 595(0.9355)

41(0.0645)

278(87.42)

39(12.26)

1(0.31)

460(0.9387)

30(0.0612)

218(88.98)

24(9.796)

3(1.22)

CYP1A2*1Fb 358(0.5581)

258(0.4188)

105(34.09)

148(48.05)

55(17.86)

256(0.5120)

244(0.4880)

73(29.2)

110(44.0)

67(26.8)

aAllele 1 is G, allele 2 is A.

bAllele 1 is A, allele 2 is C.

Power for genotypic test for CYP1A2*1C is 0.210; for CYP1A2*1F is 0.2595.


65

www.nature.com/tpj

case of CYP1A2*1C the number of individuals in each of thegenotypic categories being low, it is difficult to come to anyreliable conclusion.

As in the case of individual SNPs, no associationwas observed between CYP1A2*1C and CYP1A2*1F haplo-types with TD. Although a high linkage disequilibrium wasobserved between the two markers, CYP1A2*1C SNP wasnot found to be associated with schizophrenia, unlikeCYP1A2*1F SNP. This may be due to the low polymorphismat CYP1A2*1C making it an uninformative marker. Inaddition, the correlation coefficient r2, which representswhether both SNPs would be associated with the disease,was very low in our study (r2¼0.05).

The observed marginal association of CYP1A2*1F (locatedat 15q21) with schizophrenia may not be totally unex-pected. Chromosome 15q13–15 has been found to beassociated with schizophrenia in genome wide scans30–32

although the criteria for linkage were fulfilled in only onestudy using a subtype Periodic Catatonia (15q15 at position35.3 cM, LOD score¼ 3.57, P¼2.6� 10�5).32 Freedmanet al33 reported positive linkage of the abnormality in P50suppression with D15S1360, a dinucleotide repeat presento120 kbp from the 50 end of the coding region of the alpha7-nicotinic receptor gene (CHRNA-7), assuming an auto-somal dominant mode of transmission (LOD score¼ 5.3,Y¼0.0, Po0.000133). This linkage has been replicated inseveral other populations including African Americans,31

Germans,32 Southern African Bantu families34 and Taiwa-nese.35 Several other studies have also pointed towards thisregion as being of significance for psychiatric disorders.Further, psychoses resembling schizophrenia in Angelmansyndrome and Prader Willi syndrome (15q11–13) havebeen reported to be due to abnormal genome imprintingand in Marfan syndrome (15q21), due to mutations in thefibrillin gene. Several genes present in this region areexpressed in the brain, including tight junction protein(TJP1) and CKTSF1B1, the human homologue of DRM/GERMLIN, a secretory granule neuroendocrine protein.

CYP1A2 is located at 15q22 with two other cytochromep450 genes (CYP1A1 and CYP11A) in the vicinity ofNicotinic receptors CHRNA3, CHRNA5, CHRNB4, localizedat 15q24. Although CYP1A2 by itself has never beenimplicated in pathogenesis of schizophrenia the neighbor-ing loci as mentioned above have all been implicated. Thepolymorphism at CYP1A2*1F we have studied may by itselfpredispose to schizophrenia or it may be in linkagedisequilibria with flanking loci. CYP1A2 mRNA has beendetected in the rat brain (cortex, cerebellum, brain stem,thalamus, hippocampus, striatum36). Its presence in braintissues probably plays a role in protecting the brain fromxenobiotics and differential activity of the gene in vivo maybe responsible for differential susceptibility. The positiveassociation observed may be due to one of these reasons.Further, either different loci might be important for diseasesusceptibility in different populations or the observedassociation might be a false positive result. However, thelatter appears highly unlikely since the neighboring locihave been linked to schizophrenia in several independent

studies and our sample size is considerably large eliminatingsuch type 1 errors. On the other hand, absence of anydetectable association of CYP1A2*1F with TD in our studymay reflect a type 2 error, caused due to small number ofsmokers among TD cases and thus reduced statistical power.Therefore, we may conclude that CYP1A2*1F is weaklyassociated with schizophrenia and CYP1A2*1C shows atrend of increased severity of TD in subjects who are smokersand have received only typical antipsychotic drugs.

MATERIALS AND METHODSSubjectsRecruitment of subjects with schizophrenia was carried outat the Department of Psychiatry, RML Hospital, New Delhi.Ethical Clearance was obtained from the respective institu-tional ethical committees at RML Hospital, New Delhi andUniversity of Delhi South Campus. The Diagnostic Inter-view for Genetic Studies (DIGS) developed by Nurnburgeret al37(http://www-grb.nimh.nih.gov/gi.html) and trans-lated into Hindi by Deshpande et al38 was used forinterviewing. Consensus diagnoses were established by twopsychiatrists/psychologists using DSM IV criteria. Inter-raterreliability between the Indian and US research staff wasexamined throughout the study (k40.8038).

Schizophrenia subjects (n¼335) diagnosed according tothe DSM IV criteria (mentioned above) were included in thisretrospective cross-sectional study. All 335 subjects wereevaluated for TD by three trained research personnel (apsychiatrist and two psychologists) and consensus diagnosisestablished using the AIMS scale. Patients with overall twomild ratings or at least one moderate or high rating in any ofthe categories on AIMS scale (a single evaluation) was usedas the threshold for diagnosis of TD, following the RDC-TD.39 The Simpson-Angus Rating Scale (SARS)40 was used foradditional assessment. Other inclusion criteria were theavailability of antipsychotic treatment history (for at least3 months). Smoking status was documented by asking theproband and family members. Only those individuals whoprovided written informed consent and who were less than60 years of age were included in this study.

ControlsAge, sex and ethnicity matched healthy individuals(n¼250) were recruited from the collaborating hospitalsand the University Health Center. The controls and theirfamilies had no previous history of any neuropsychiatricdisorder.

Genetic AnalysisPeripheral venous blood (10 ml) was drawn from allparticipating subjects and used for DNA isolation. GenomicDNA was isolated using the conventional phenol chloro-form extraction method.41 Polymerase chain reaction wasused for the amplification of specific regions of CYP1A2gene (AH002667). Information on SNPs to be evaluated inthis study was obtained from the CYPallele nomenclatureweb site www.imm.ki.se/CYPalleles. Six sets of primers toamplify the specific regions encompassing the SNPs to be


66


evaluated were designed using Primer ver 1.0. Of the sixSNPs, four (CYP1A2*2, *4, *5, *6) were evaluated by SnaP-shot method (Table 7) using commercial facility at LabIndia,New Delhi, India. Two SNPs (CYP1A2*1C and CYP1A2*1F)were assayed by PCR-RFLP approach. Primers used foramplification of CYP1A2*1C were F50-GCT ACA CAT GATCGA GCT ATA C-30 and R50-CAG GTC TCT TCA CTG TAAAGT TA-30.17 Briefly, the PCR was performed in a totalreaction volume of 20ml containing 100 ng of DNAtemplate, 200 mM dNTPs, 0.5mM of each primer, 0.5 U ofTaq polymerase (Roche) and 1�PCR reaction buffer.Cycling conditions were initial denaturation at 951C for5 min followed by 35 cycles of 951C for 45 s, 561C for 45 s,721C for 1 min and final extension at 721C for 7 min. Theresulting 596-bp segment was digested with DdeI andresolved on a 3% ethidium bromide containing agarosegel. The wild-type allele gave an undigested band of 596 bp,and the digested product was resolved into 464 and 132 bpband. The 520-bp region in intron I containing CYP1A2*1Fwas amplified using the primers F 50-TGG AGT GGT CACTTG CCT CT-30 and R 50-CTG GCT CAT CCT TGACAG T-30.The cycling conditions were initial denaturation at 951C for5 min, followed by 35 cycles of 951C for 30 s, 581C for 30 s,721C for 45 s and final extension at 721C for 5 min. The PCRproducts were digested with 1 U of Bsp120I restrictionendonuclease. The allele A showed as an undigested bandat 520 bp, the presence of allele C resulted in a 373 and a147 bp band. The digestion products were resolved on a 3%ethidium bromide containing agarose gel. Complete diges-tion was observed in all the samples for both the SNPs. In

addition, samples with known genotypes were included ineach set of digestion to ensure that the observed genotypeswere not due to partial/incomplete digestion.

SnaPshotPCR

PCR reactions were performed with approximately 100 ngof genomic DNA. Briefly, 25 ml reactions were set up with1.5 mM MgCl2, 100 mM of each of dATP, dCTP, dGTP anddTTP, 0.5 U of AmpliTaq Gold (ABI) and 20 pmol each offorward and reverse primers. The cycling conditions wereinitial denaturation at 961C for 13 min followed by 40 cyclesof denaturation 951C for 10 s, annealing and amplificationat 601C for 1 min. The final extension was carried out at721C for 30 min.

SnaPshot reactions

For removal of excess PCR primers and dNTPs, the PCRproducts were digested with 0.25 U of Exonuclease I(Epicenter) and 0.5 U of Shrimp Alkaline Phosphatase (MBIFermentas) at 371C for 2 h, followed by inactivation ofExonuclease I and SAP at 801C for 15 min. Subsequently,SnaPshot reactions were set up using 2 ml of digested PCRproducts, 1.25 ml of SnaPshot ready reaction mix (ABI), 0.5 mlof 5� sequencing Buffer, 0.25 mM concentration of SnaPshotprimer (0.25 uM of each primer in case of multiplexreactions) and MilliQ water to make up the reaction volumeto 5 ml. The cycling conditions were 25 cycles at 961C for10 s, 501C for 5 s and 601C for 30 s.

Table 7 Primer sequences for PCR amplification and sequences of the interrogation probes (for SnaPshot reactions) for fourCYP1A2 SNPs

SNP Primer/PROBE sequence

CYP1A2*2 F50-TCA GTG TTT ATC AAA ATG ACT GAG GAA-30 Locus amplificationR50-TCA TCC TTG ACA GTG CCA GGT-30

IF50-GAT CTT TTC CTG GCC TCT GCC ATC TT-30 SNP interrogationIR50-GAT CAA AAG CAC CCA GAA TAC CAG GCA-30

CYP1A2*4 F50-TTA TGT GCC TGC TGT GTG CA-30 Locus amplificationR50-TCA CAG GTG GAA CTT TCT CCT GA-30

IF50-GAT CAA AAA AAA AAA AAA AAT CCT CCT TCT TGC CCT TCA-30 SNP interrogationIR50-GAT CAA AAA AAA AAA AAA AAA GGC CTC ACC TGT GGG GGA-30

CYP1A2*5 F50-TTC CAG CCC TGA GCC TCA-30 Locus amplificationR50-CAT CCT AGT TGA TTC CTG TGC ACT-30

IF5-GAT CAC ATC CCC AAG AAA TGC T-30 SNP interrogationIR50-CTA GCC ACT GGT TTA CGA AGA CA-30

CYP1A2*6 F50-TTC CTT CCC ACC TAC CCT TCA-30 Locus amplificationR50-GCT GTA GCA GGA TGG CCA-30

IF50-GAT CTT TTT TTT TTT TTT CTG AGT TCC GGC CTG AG-30 SNP interrogationIR50-GAT CAA AAA AAA AAA AAC ATC GGC GGT GAG GAA CC-30


67

www.nature.com/tpj

After the reactions were completed, 0.5 U of Calf IntestinalAlkaline Phosphatase (MBI Fermentas) was added to eachreaction and incubated at 371C for 1 h, followed byinactivation of CIP at 801C for 15 min. For running theproducts in ABI 3100 Genetic Analyzer, 2 ml of each ofSnaPshot products (for multiplex injection) were mixedwith 8 ml of Hi Di Formamide (ABI), denatured at 951C for5 min and snapchilled on ice. The samples were thenanalyzed using Run Module SNP22_POP4, Dye set E5,Pop4 polymer and 22 cm Capillary Arrays.

Data analysis

The data were first analyzed with GeneScan Analysis Soft-ware 3.7. Subsequently, the GeneScan data were analyzedwith Genotyper Software 3.7 for automated allele callingand generation of tables.

Statistical AnalysisFrequencies of each of the SNPs evaluated in this study werecalculated for both the healthy volunteers as well as subjectswith schizophrenia. Hardy–Weinberg equilibrium was testedfor the observed genotypes using w2 test. The genderdistribution was compared using w2 test and the mean ageamong different groups such as TD positive (TD-Y), TDnegative (TD-N) and controls was compared using t-test.Association of individual SNPs was tested using a w2 test. Allstatistical tests were carried out using the Statistical Packagefor Social Sciences (SPSS) version 11.0. The mean AIMS scorewas compared between each of the genotypic classes usinganalysis of covariance (ANCOVA), with age as a covariate.Levene test for homogeneity of variance was carried out tocheck the assumption of equal variance among eachgenotypic class, normal distribution of genotypic categorieswas tested using Kolmogrov–Smirnov Z test. Power calcula-tion was carried out based on mean AIMS score and samplesizes, with alpha set to 0.05. In case of violation of any of theassumptions of ANCOVA Kruksal Wallis H test was per-formed. Power calculations were also carried out for case–control analysis based on differences in the allele frequencybetween cases and controls and sample size [PAWE42,43].Haplotype frequencies were calculated using PHASEver2.0.244,45 and haplotypic association was tested usingCLUMP.46 Linkage disequilibrium between the markers wascalculated using EMLD (www.linkage.rockefeller.edu).

ACKNOWLEDGEMENTSThis work is supported by Indo-Israel grants BT/IC-2/Israel/Deshpande/2002 and BT/IC-2/00/Smita/99 (to SND, BKT, BL);Department of Biotechnology (Govt of India) Grant BT/PR2425/Med 13/089/2001 (to BKT and SND); Indo-US Grant N-443-645 (toVLN, BKT, SND); and a research fellowship from the UniversityGrants Commission, New Delhi to AKT.

DUALITY OF INTERESTThe authors declare that they have no competing financialinterests.

REFERENCES1 Yassa R, Ananth J. Familial Tardive dyskinesia. Am J Psychiatry 1981; 138:

1618–1619.2 Muller DJ, Schulze TG, Knapp M, Held T, Krauss H, Weber T et al.

Familial occurrence of Tardive dyskinesia. Acta Psychiatr Scand 2001;104: 375–379.

3 Kane JM, Smith JM. Tardive Dyskinesia: prevalence and risk factors 1959to 1979. Arch Gen Pscychiaty 1982; 39: 473–481.

4 Bergen J, Kitchin R, Berry G. Predictors of the course of TardiveDyskinesia in patients receiving neuroleptics. Biol Psychiatry 1992; 32:580–594.

5 Jeste DW, Wyatt RJ. Dogma disputed: is tardive dyskinesia due topostsynaptic dopamine receptor sensitivity? J Clin Psychiatry 1981; 42:455–457.

6 Casey DE, Geralch J, Bjondal N. Levodopa and receptor sensitivitymodification in tardive dyskinesia. Psychopharmacology 1982; 78:89–92.

7 Geralch J, Casey DE. Sulpiride in tardive dyskinesia. Acta Psychiatr Scand1984; 311: 93–102.

8 Mehsul CK, Casey DE. Regional, reversible ultrastructural changesin rat brain with chronic neuroleptic treatment. Brain Res 1989; 489:338–346.

9 Gordon JH, Fields JZ. A permanent dopamine receptor upregulationin the ovariectomized rat. Pharmacol Biochem Behav 1989; 33:123–125.

10 Basile VS, Masellis M, Potkin SG, Kennedy JL. Pharmacogenomics inschizophrenia: the quest for individualized therapy. Hum Mol Genet2002; 11: 2517–2530.

11 Lerer B, Segman RH, Fangerau H, Daly AK, Basile VS, Cavallaro R et al.Pharmacogenetics of Tardive Dyskinesia: combined analysis of 780patients supports association with dopamine D3 receptor gene Ser9Glypolymorphism. Neuropsychopharmacology 2002; 27: 105–118.

12 Shimada T, Yamazaki H, Mimura M, Inui Y, Gungerich FP. Interindivi-dual variations in human liver cytochrome P450 enzymes involved inthe oxidation of drugs, carcinogens and toxic chemicals: studies withliver microsomes of 30 Japanese and 30 Caucasians. J Pharmacol Ther1994; 270: 414–423.

13 Arthur H, Dahl ML, Siwers B, Sjoqvist F. Polymorphic drug metabolismin schizophrenia patients with tardive dyskinesia. J Clin Psychpharmacol1995; 15: 211–216.

14 Armstrong M, Daly AK, Blennerhassett R, Ferrier N, Idle JR. Anti-psychotic induced movement disorders in schizophrenics in relation toCYP2D6 genotype. Br J Psychiatry 1997; 170: 23–26.

15 Basile VS, Ozdemir V, Masellis M, Walker ML, Meltzer HY, Lieberman JAet al. A functional polymorphism of the cytochrome P450 1A2(CYP1A2)gene: association with tardive dyskinesia in schizophrenia. MolPsychiatry 2000; 5: 410–417.

16 Smith SS, Fiore MC. The epidemiology of tobacco use, dependence andcessation in the United States. Primary Care 1999; 26: 433–461.

17 Nakajima M, Yokoi T, Mizutani M, Kinoshita M, Funayama M, KamakatiT. Genetic polymorphism in the 50-flanking region of human CYP1A2gene: effect on the CYP1A2 inducibility in humans. J Biochem 1999;125: 803–808.

18 Sachse C, Brockmoller J, Bauer S, Roots I. Functional significance of aC4A polymorphism in intron I of the cytochrome P450 CYP1A2 genetested with caffeine. Br J Clin Pharmacol 1999; 47: 445–449.

19 Guy W (ed). Early Clinical Drug Evaluation Unit b Assessment Manual.US Department of health and human services, National Institute ofMental Health (NIMH): Rockville, Maryland 1976.

20 Schultze TG, Schumacher J, Muller DJ, Krauss H, Alfter D, Maroldt Aet al. Lack of association between a functional polymorphism of thecytochrome P450 1A2 (CYP1A2) gene and tardive dyskinesia inschizophrenia. Am J Med Genet 2001; 105: 498–501.

21 Chong S, Tan E, Tan CH, Mythily. Smoking and Tardive Dyskinesia: lackof involvement of the CYP1A2 gene. J Psychiatry Neurosci 2003; 28:185–189.

22 Basile VS, Masellis M, Badri F, Paterson AD, Meltzer HY, Lieberman JAet al. Association of the MscI polymorphism of the dopamine D3receptor gene with tardive dyskinesia in schizophrenia. Neuropsycho-pharmacology 1999; 21: 17–27.

23 Kane JM. Clinical efficacy of clozapine in treatment-refractory schizo-phrenia: an overview. Br J Psychiatry 1992; 160: 41–45.


68


24 Yassa R, Jeste DV. Gender differences in tardive dyskinesia: a criticalreview of the literature. Schizophr Bull 1992; 18: 701–715.

25 Chevalier D, Cauffiez C, Allorge D, Lo-Guidice JM, Lhermitte M, Lafitte JJet al. Five novel natural allelic variants – 951 A4C, 1042 G4A (D348N),1156A4T (I386F), 1217 G4A (C406Y) and 1291 C4T (C4341Y) – ofthe mutation in a French Caucasian population. Hum Mutat 2001; 17:355–360.

26 Nebert DW. Polymorphism in drug metabolizing enzymes: what is theirclinical relevance and why do they exists? Am J Hum Genet 1997; 60:265–271.

27 Ikeya K, Jaiswal AK, Owens RA, Jones JE, Nebert DW, Kimura S. HumanCYP1A2: sequence, gene structure, comparison with mouse and ratorthologous gene, and differences in liver mRNA expression. MolEndocrinol 1989; 3: 1399–1408.

28 Pickwell GV, Shih H, Quattrochi LC. Interaction of upstream stimulatoryfactor proteins with an E-box located within the human CYP1A2 5’-flanking gene contributes to basal transcriptional gene activation.Biochem Pharmacol 2003; 65: 1087–1096.

29 Hamdy SI, Hiratsuka M, Narahara K, Endo N, El-Enany M, Moursi Net al. Genotyping of four genetic polymorphisms in the CYP1A2 genein the Egyptian population. Br J Clin Pharmacol 2003; 55: 321–324.

30 Coon H, Jensen S, Holik J, Hoff M, Myles-Worsley M, Reimherr F et al.Genomic scan for genes predisposing to schizophrenia. Am J Med Genet1994; 54: 59–71.

31 Kaufman CA, Suarez B, Malaspina D, Pepel J, Svarakic D, Meyer J et al.NIMH genetics initiative millenium schizophrenia consortium: linkageanalysis of African American pedigrees. Am J Med Genet 1998; 88:29–33.

32 Stober G, Saar K, Ruschendorf F, Meyer J, Nurnberg G, Jatzke S et al.Splitting schizophrenia: periodic catatonia - susceptibility locus onchromosome 15q15. Am J Hum Genet 2000; 67: 1201–1207.

33 Freedman A, Coon H, Myles-Worsley M, Orr-Urtreger A, Olincy A, DavisA et al. Linkage of a neurophysiological deficit in schizophrenia to achromosome 15 locus. PNAS 1997; 94: 587–592.

34 Riley B P, Makoff A, Mogudi-Carter M, Jenkins T, Williamson RD, MurrayR. Haplotype transmission disequilibrium and evidence for linkage ofthe CHRNA7 gene region to schizophrenia in Southern Africal Bantufamilies. Am J Med Genet 2000; 96: 196–201.

35 Liu C, Hwo H, Lin M, Ou-yang W, Lee S, Fann C et al. Suggestiveevidence for linkage of schizophrenia markers at chromosome 15q13–14 in Taiwanese families. Am J Med Genet 2000; 105: 658–661.

36 Morse DC, Steim AP, Thomas PE, Lowndess HE. Distribution andinduction of cytochrome P450 1A1 and 1A2 in rat brain. Toxicol ApplPharmacol 1998; 152: 232–239.

37 Nurnberger Jr JI, Blehar MC, Kaufmann CA, York-Cooler C, Simpson SG,Harkavy-Friedman J et al. Diagnostic interview for genetic studies.Rationale, unique features, and training. NIMH Genetics Initiative. ArchGen Psychiatry 1994; 5: 849–859.

38 Deshpande SN, Mathur MNL, Das SK, Bhatia T, Sharma SD,Nimgaonkar VL. A hindi version of the diagnostic interview for geneticstudies. Schizophr Bull 1998; 24: 489–493.

39 Schooler NR, Kane JM. Research diagnoses for tardive Dyskinesia. ArchGen Prychiatry 1982; 39: 486–487.

40 Simpson GM, Angus JWS. Drug-induced extrapyramidal disorder – arating scale for extrapyramidal side effects. Acta Psychiatr Scand Suppl1970; 212: 20–27.

41 Sambrook J, Fritsch EE, Maniatis T. Molecular Cloning: A LaboratoryManual. 2nd edn, Cold Spring Harbor Laboratory press: Cold SpringHarbor, NY, 1989.

42 Gordon D, Finch SJ, Nothnagel M, Ott J. Power and sample sizecalculations for case–control genetic association tests when errorspresent: application to single nucleotide polymorphisms. Hum Heredity2002; 54: 22–33.

43 Gordon D, Levenstien MA, Finch SJ, Ott J. Errors and linkagedisequilibrium interact multiplicatively when computing sample sizes forgenetic case–control association studies. Pacific Symposium on Biocom-puting. 2003, pp 490–501.

44 Stephens M, Smith NJ, Donnelly P. A new statistical method forhaplotype reconstruction from population data. Am J Hum Genet 2001;68: 978–989.

45 Stephens M, Donnelly P. Comparison of Bayesian methods forhaplotype reconstruction from population genotype data. Am J HumGenet 2003; 73: 1162–1169.

46 Sham PC, Curtis D. Monte Carlo tests for associations between diseaseand alleles at highly polymorphic loci. Ann Hum Genet 1995; 59:97–105.


69

www.nature.com/tpj

Genetic susceptibility to Tardive Dyskinesia in chronic schizophrenia subjects: V. Association of CYP1A2 1545 C>T polymorphism

Documents